Coverage evaluations in Tunnels

A4 RZ COVERAGE orange 3.FE

08.07.1999 9:55 Uhr

Seite 2

RADIO FREQUENCY SYSTEMS

COVERAGE EVALUATIONS IN TUNNELS APPLYING RADIATING CABLES

Reprint from the proceedings of the ITC Conference Amsterdam, March 1997 RFS kabelmetal H.-D. Hettstedt, M. Davies, B. Herbig, R. Nagel

R F S

C o n n e c t

w i t h

t h e

b e s t

®

Innens. A4 -Orange 1-3

R F S

08.07.1999 9:49 Uhr

Seite 1

C o n n e c t

w i t h

t h e

b e s t

®

Text COVERAGE EVALUA….FE

08.07.1999 9:54 Uhr

Seite 1


ABSTRACT

2. COVERAGE MODELS

This paper analyses coverage predictions in tunnels using radiating cables. It is demonstrated that cable data gained from free space measurements results are applicable to tunnel environments. Fading models used for radio communication in free space applied to tunnels explain effects of electrical behaviour of radiating cables in tunnels. A coverage prediction is performed by system loss calculations based on cable data. The coverage model for tunnels is similar to that of free space. Measurement results are presented from a typically equipped metro tunnel considering a rectangular concrete section and another consisting of a combination of a concrete section with a steel tube. The results show concurrence between theory and practice.

1. INTRODUCTION

For radio coverage in free space, two types of coverage are important for the system design, the area coverage and the contour coverage. Assuming a specific fading characteristic, e.g. Rayleigh fading or Ricean fading, the area coverage can be computed from the coverage measured on any contour surrounding this area. In a tunnel along a radiating cable the coverage can be computed directly from measurement results of the system loss. A normal system design approach is to predict the reception probability at the end of the cable section using known cable data: cable loss and coupling loss. The reception probability thus gained is comparable with the contour coverage in free space. Of special interest are the influences of the tunnel surroundings on the coverage characteristics and comparison between predicted and measured coverage values. This papers helps to clarify this complex. A further interesting point relevant to system design is that of matching the coverage in free space with that in tunnels. Fading models in tunnels are investigated for the case of radiating cables. Figure 2.1: Sketch of a Free Space Scenario

In order to meet specific requirements of system reliability, radio coverage normally has to be confirmed theoretically and by measurements. In free space applications, the situation and design procedure is well known. In tunnels the situation is very different, though the measurement conditions are actually simpler. Normally a tunnel radio system is a portion of a larger radio system supplying both free space and tunnels. So the coverage in both areas is of special interest as well as the interfaces between them. Therefore both situations are considered in this paper. 2.1 Free Space In free space, coverage requirements are defined by a specific minimum signal reception level within a defined percentage of an area. For public safety e.g. an area coverage of 98 % is needed. FIG. 2.1 shows an area in a simplified form within a circular contour enclosing a Base Station antenna. As the dimensions can be of the order of several kilometres, the difficulty of confirmation by direct measurements is obvious. So the procedure of measuring only the contour coverage is a useful simplification, but still involves considerable effort. A mathematical relationship between area and contour coverage is given in [1] assuming Rayleigh fading under these specific conditions. The area in FIG. 2.1 is sepa-

Wood Area Antenna

Urban Area

Hilly Area

Plane Area

Circular Contour

1


08.07.1999 9:54 Uhr

Seite 2


rated into 4 different segments representing very different conditions for radio wave propagation. This extreme inhomogenity is to demonstrate the complexity of exact coverage evaluation in large areas. 2.2 Tunnels Confirmation of radio coverage in tunnels is easier in practice because the space is clearly limited to a narrow area which is normally directly available for measurements. So the results of field strength measurements along the radiating cable can be used to evaluate reception probability and coverage evaluation. FIG. 2.2 shows the situation inside a tunnel crosssection giving an image that signal reflections from the tunnel wall are an important factor. To gain theoretical coverage predictions as part of the system design as a whole, the conditions for electromagnetic propagation in the tunnel must be analysed carefully.

• • • •

Measurement results for specific antenna heights and distances from the cable should be valid within the whole cross-section. Cable parameters gained from free space measurements must be transferable to tunnel applications. Tunnel influences must be clear and predictable. The characteristics of the signal strength variations along the cable must be analysed mathematically.

Multipath Radiating Cable Direct Path

3. FADING MODELS

Signal transmissions generally show field strength variations depending on local characteristics. These phenomena, known as fading, must all be considered in evaluating radio communication parameters. In order to obtain comparable procedures, conditions in free space and in tunnels must be analysed. 3.1 Fading in Free Space In free space, the fading effects can be separated into two parts, a long and a short-term fading, see e.g. [1]. The long-term fading represents signal strength variations due to specific local attenuation and blocking effects additionally to the normal attenuation of e.m. propagation over distance. The name is due to its nearly constant characteristic on time. The statistical distribution derives from a lognormal function with a standard deviation of typically 5.5 dB for frequencies up to 1 GHz. Superposed on this effect there is a short-term fading resulting from multipath propagation with a high density of signal variations. The statistical distribution is related to Rice or Rayleigh functions, see e.g. [1]. Rice fading is typical when more than 50% of the signals are propagated on the direct path. The Rayleigh function must be applied when multipath propagation predominates. FIG. 3.1 shows these relationships resulting in a free space attenuation at a specific distance to a BTS antenna. Moving along the circular contour in FIG. 2.1, we obtain both lognormal fading over a constant attenuation D due to the area specific characteristics as well as multipath effected fading. In case of Rayleigh fading the mean value will have a lognormal distribution. In this case the mean value differs from the median (i.e. 50% reception probability).

Mobile Tunnel Cross-Section 2

Figure 2.2: Sketch of a Tunnel Scenario


08.07.1999 9:54 Uhr

Seite 3


Attenuation

Lognormal Fading

Short-term Fading

Free Space Attenuation

e Figure 3.1: Fading Effects in Free Space

Direct Path Dominating

Multipath Dominating

Distance (Log. Scale)

Signal Levels

(ca. 5dB) (ca. 12dB for 98% Coverage)

Total Margin

Margin for Lognormal Fading Margin for Multipath Mading

Fading Margin (e.g. 12dB SINAD, 50% Coverage) Static C/N

(C/N = 1) Inferred Noise Level

Noise Figure

Thermal Noise Level e.g. Frequency

Figure 3.2: Fading Margins

Each of these effects must be considered in system calculations, see FIG. 3.2. In order to achieve a required coverage, a fading margin is needed which represents both the long-term and short-term fading. In [1] the mathematical relationship between contour and area coverage is given. A typical calculation resulting from Rayleigh fading is: A 95% contour coverage requires a 12 dB short-term margin and results in a 97% area coverage.

3.2 Fading in Tunnels In tunnels a line of sight to the radiating cable can normally be assumed. The conditions for Rice fading are evidently given. However, the situation is complicated by the cable itself. There is no discrete radiating source, instead the cable has a function of a distributed antenna. This leads to an electrical field along the cable with typical interference from interactions of different types of e.m. waves generated from the cable. For a detailed description of the function of radiating cables see [2].

3


08.07.1999 9:54 Uhr

Seite 4


4

Figure 3.3: Measurement Results of Coupled Mode Cables in Free Space and Tunnel


08.07.1999 9:54 Uhr

Seite 5


Figure 3.4: Measurement Results of Radiating Mode Cables in Free Space and Tunnel

5


08.07.1999 9:54 Uhr

Seite 6


So radiating cables show fading characteristics even without tunnel environments. They can be classified into two general types, operating either in the coupled mode or the radiating mode. Both types show different field characteristics. FIG. 3.3 shows measurement results of the coupling loss of an RLF cable operating in the coupled mode at 960 MHz. The upper diagram shows results gained from free space measurements under standard conditions. The lower diagram shows results from tunnel measurements under equivalent conditions. In FIG. 3.4 similar results are shown for a RAY type cable operating in the radiating mode.

Figure 3.5: Probability Density Functions for Cables in Free Space and Tunnel

6

Both types of cables show similar statistical distributions of the coupling loss for free space and tunnel environments. This means that multipath in tunnels within the cross-section is of secondary order only and that the cable characteristics are clearly dominated by their functions. These are results of a more exhaustive investigation of different types of cables, see [3]. [3] also shows that the same cables give results in free space and tunnels which can differ up to ± 5 dB, although, at discrete frequencies, the 5%, 50% and 95% reception probability values relative to each other, remain the same.

Upper Part: Coupled Mode Cable, Lower Part: Radiating Mode Cable


08.07.1999 9:54 Uhr

Seite 7


The statistical distributions of field strength in free space and tunnels are thus apparently equivalent, but the tunnel environments lead to an offset of max. ± 5 dB. These results would suggest that the tunnel environments result in a type of fading comparable with the lognormal fading in free space. A thorough analysis of the distribution of this effect would require a very great number of tunnel measurements under otherwise unchanged conditions. The probability density function of the cable itself, whether in free space or in tunnels, does not fulfil the parameters of the Rayleigh or the Ricean characteristic. The mean value of a Rayleigh distribution is 1.25 times the standard deviation and that of a Ricean distribution is even higher (see [1]), whilst graphical and numerical analysis of radiating cable coupling loss shows the mean value to be actually less than that of the standard deviation (typically 90%). This effect is only possible when a high degree of asymmetry is present in the distribution (remembering that negative values for coupling loss in non-logarithmic terms are not possible). The asymmetry apparent in the PDF’s shown in FIG. 3.5 is characterised by a very steep incline towards zero (compared to Rice or Rayleigh distributions) and a very gradual decline towards higher signal strengths. This tendency is typical of logarithmic scaling, so that evidently a lognormal element predominates to an extent that other superimposed characteristics are no longer recognisable. We can thus suppose that the radiating cable pattern contains both a Rayleigh and a lognormal component which would mean that the resulting distribution is then a Suzuki distribution (see [1]). Further research is necessary to verify this, deriving the appropriate parameters for a hypothesis and subjecting it to a chi squared test, and is the subject of present research.

4. COVERAGE PREDICTIONS IN TUNNELS

For the coverage prediction in tunnels an approach is useful which accords with that for free space. The system loss for the end section of a long radiating cable can be computed using cable data gained from free space measurements. The considerations of the previous chapters show this procedure to be correct. This computation leads to a coverage comparable with the contour coverage in free space and is the basis for the overall coverage along the whole cable. Fig. 4.1 shows the relationship between cable loss, coupling loss, system loss, margin for „lognormal“ fading and the min. reception level. Using the cable data from standard tests the length of the end-section is of course the same as of the test length, normally 100 m. This procedure of system loss computations leads to situations where the min. reception level at the end of the cable always meets the coverage requirement. This is a desirable result for system design. The difference between the coverage along the whole cable and that of the end-section depends on the amount of total cable loss. Fig. 4.2 shows the system loss diagrams of an RLF type cable for a 500 m and a 1000 m section, theoretically evaluated. A comparison of the reception probability curves show that the distribution is stretched by increasing the cable losses due to the double cable length. The differences between the 50%- and 95%-values are remarkable. The divergence of end-section coverage and area coverage is shown by a brief analysis: Cable type: Frequency: Cable loss: Coupling loss:

End-Section Coverage Length [m] 100 500 1000

RLF 13/33-1800 900 MHz 3.4 dB/100 m 71.3 dB/50%, 82.5 dB/95%

Whole-Section Coverage

Sys. loss (50%)

Sys. loss (95%)

Sys. loss (50%)

74.7 dB 88.3 dB 105.3 dB

85.9 dB 99.5 dB 116.5 dB

80.3 dB 89.1 dB

Sys. loss (95%) 93.4 dB 106.9 dB

Sys. loss (98%)

97.3 dB 111.1 dB

7


08.07.1999 9:54 Uhr

Seite 8


This analysis shows that the procedure of computing the system loss for a 95% coverage of the end-section leads to an area coverage of more than 98%. The differences depend on cable types, lengths and frequencies. The coverage prediction has to be performed analytically case by case.

System Loss

Cable Loss

Lognormal Fading

95%

Coupling Loss Lognormal Margin

Min. Reception Level (C/N= 12dB) 100m

Distance (Linear Scale)

Figure 4.1: Diagram of System Loss in Tunnels

5. MEASUREMENT RESULTS In a longer section of the UESTRA metro tunnels in Hanover, different types of radiating cables were tested. The cables are installed on side walls at train window level in a single bore in one direction. The tunnels are mostly of concrete material and of rectangular size including niches, side changes of the cabling and diverse typical discontinuities. These environments offer the opportunity for tests under typical installation conditions.

8

Two representative test results were chosen. FIG. 5.1 shows in the upper part a diagram of system loss measured in a rectangular tunnel section between two metro stations. The diagram below shows the system loss resulting from measurements in a tunnel section where two different types are combined: a concrete rectangular tunnel with a metal tube. Both diagrams are related to cable lengths of appr. 500 m. For both tests the same cable type RLF 17/44 was used for the tests at 960 MHz.

The test antenna was fixed outside the train in the centre of the front window at vertical orientation. So the test conditions were equal. In the upper diagram for the system loss in the rectangular tunnel it can be seen that the slope shows small variations which can be explained by discontinuities in the tunnel and local additional cable losses from connecting jumper cables. The analysis of the extracted coupling loss shows that there is a nearly constant offset between the free space and tunnel values which are within the ± 5 dB variation due to the expected „lognormal“ fading effect. Comparing the measured system loss with the calculated one it can be seen that the 50% values fit very well. The predicted 95% value is again equivalent with the measured 98% value. The difference between the 5% values are comparable with the difference between


08.07.1999 9:54 Uhr

Seite 9


Figure 4.2: Theoretical System Loss Results Extrapolated from Measurements on a Cable of 150m

9


08.07.1999 9:54 Uhr

Seite 10


10

Figure 5.1: System Loss Results Measured in a Concrete Tunnel and in a Tunnel of a Combination of a Concrete Section with a Steel Tube


08.07.1999 9:54 Uhr

Seite 11


Figure 5.2: Comparison of Coupling Loss Results Gained in Sections of a Concrete Tunnel and of a Steel Tube

11


08.07.1999 9:54 Uhr

Seite 12


Coupling Loss Reception Probability 5% 50% 95%

Free Space Data

65.7 dB 72.9 dB 84.6 dB

Tunnel Data

69.3 dB 76.7 dB 88.0 dB

the 95% values, see the table below. In conclusion, this example of practical application shows that the predicted coverage of 95% at the far end of the cable leads to a total coverage of 98%. Further interesting results were gained from the measurements in the section where two very different types of tunnels are connected directly. As can be seen from the lower diagram in FIG. 5.1 there is a step of appr. 10 dB in system loss at the junction from the concrete tunnel section to the steel tube. This means there is a remarkable difference between the coupling losses in the different sections. In FIG. 5.2 the coupling loss characteristics for both tunnel sections are shown separately. The upper diagram shows the coupling loss for the concrete section, the lower one that for the first 130 m of the steel tube. Direct comparison demonstrates an effective lower coupling loss in the tube at a difference of appr. 10 dB and a more regular fading characteristic. The reason for this effect is of course higher reflections resulting in lower losses. Another interesting result is that there is no change in the distribution function for the tube. It can be assumed that even under highly reflective environments the cable characteristics do not change and that a constant offset based on typical influences of a tunnel can again be observed, in this case with positive results.

6. CONCLUSIONS

12

It was demonstrated that distribution characteristics of radiating cables in tunnels are equivalent to those in free space. Differences of electrical behaviour can be explained by fading models from free space. A coverage

System Loss Offset

Predicted from Tunnel Data

Measured Data

+ 3.4 dB + 3.8 dB + 3.4 dB

73.4 dB 88.2 dB 106.9 dB

77.5 dB 88.4 dB 102.0 dB

prediction for tunnel sections supplied by radiating cables can be performed from system loss calculations based on cable data. Measurement results of a metro tunnel in typical concrete environments and in a highly reflective steel tube confirm theoretical calculations made in free space, demonstrating the independence of the cable’s behaviour to the environment.

7. ACKNOWLEDGEMENTS

The authors would like to extend their thanks to the UESTRA AG, Hanover, for their kind permission to use the tunnels and to Mr. Witte and Mr. Reuter for their good collaboration in forms of advice and supply. Furthermore, they wish to thank Mr. Mahlandt from the RFS Cable Development Department for his helpful advice and for the supply of software.

8. REFERENCES

[1] M. D. Yacoub: Foundations of Mobile Radio Engineering, CRC Press, 1993 [2] H.-D. Hettstedt: Development and Applications of Leaky Feeders, International Seminar on Communications Systems For Tunnels, London, 1993 [3] H.-D. Hettstedt, B. Herbig, G. Klauke, R. Nagel: Comparison of Performances of different Leaky Feeders in a Metro Tunnel, Tunnel Control & Communication, Basel, 1994


08.07.1999 9:54 Uhr

Seite 13


13

A4 RZ COVERAGE orange 3.FE

R

F

08.07.1999 9:55 Uhr

S

G

R

Seite 1

O

U

P

M

E

M

B

E

R

S

GERMANY: RFS kabelmetal Tel: +49-511 676 2520 Fax: +49-511 676 2521

BRAZIL: RFS Brasil - kmP Tel: +55-11 7961 2433 Fax: +55-11 494 2937

USA: RFS Cablewave Tel: +1-203 630 3311 Fax: +1-203 821 3852

AUSTRALIA: RFS Australia Tel: +61-3 9761 5700 Fax: +61-3 9761 5711

FRANCE: RFS France Tel: +33-1 3423 6200 Fax: +33-1 3423 6324

USA: RFS Cablewave Eastern Region Tel: +1-203 630 3311 Fax: +1-203 821 3852 USA: RFS Cablewave Midwest Region Tel: +1-630 516 0106 Fax: +1-630 516 0207 USA: RFS Cablewave Western Region Tel: +1-510 661 9620 Fax: +1-510 661 9630 USA: RFS Cablewave Southeast Region Tel: +1-770 649 9929 Fax: +1-770 649 8884

UK: RFS UK Tel: +44-1494 447 110 Fax: +44-1494 442 742

SINGAPORE: RFS Singapore Tel: +65-272 9233 Fax: +65-272 9266

CHINA: RFS Beijing Tel: +86-10 659 00370 Fax: +86-10 659 06932

ITALY: RFS Italia Tel: +39-039 2 14 85 69 Fax: +39-039 2 14 85 53

THAILAND: RFS Thailand Tel: +66-2 273 8051-2 Fax: +66-2 273 8053

CHINA: RFS Shanghai Tel: +86-21 5836 8417 Fax: +86-21 5836 8291

RUSSIA: RFS Moscow Tel: +7-095 258 0632 Fax: +7-095 258 0633

INDONESIA: RFS Indonesia Tel: +62-21 797 60 76 Fax: +62-21 797 60 77

HONG KONG: RFS Hong Kong Tel: +852-2861 4438 Fax: +852-2861 4488

S

A

L

E

S

O

F

F

I

C

E

S

Please visit us on the internet at http: //www.rfs-group.com 2299.050.PM 171 001 56 - 01

R F S

C o n n

Coverage evaluations in Tunnels

Recommend Documents